1,4,5,8,9,12-Hexamethyltriphenylene. a molecule with a flipping twist

Article abstract of DOI:10.1021/ja074120j

The synthesis and characterization of 1,4,5,8,9,12-hexamethyltriphenylene (5) is described. In the solid state, X-ray crystallographic studies reveal that compound 5 presents a highly distorted C₂ geometry with a 53° end-to-end twist. In soluti

Full text of DOI:10.1021/ja074120j

Published on Web 10/06/2007
1,4,5,8,9,12-Hexamethyltriphenylene. A Molecule with a
Flipping Twist
Yi Wang, Andrew D Stretton, Mark C McConnell, Peter A. Wood, Simon Parsons,
John B Henry, Andrew R Mount,* and Trent H Galow*
Contribution from the School of Chemistry, UniVersity of Edinburgh, The King’s Buildings,
West Mains Road, Edinburgh EH9 3JJ, U.K.
Received June 6, 2007; E-mail: trent.galow@ed.ac.uk
Abstract: The synthesis and characterization of 1,4,5,8,9,12-hexamethyltriphenylene (5) is described. In
the solid state, X-ray crystallographic studies reveal that compound 5 presents a highly distorted C2 geometry
with a 53° end-to-end twist. In solution, variable-temperature 1H NMR studies and molecular modeling
present a story of rapid dynamic conformational interconversions between two C2 enantiomers (with a low
activation barrier) and a slower C2-D3 interconversion (with a relatively high barrier)sthe first time clear
evidence of conformational interchange for these hindered triphenylenes has been provided. Further studies
have established that 5 is a fluorescent stable blue emitter, and that the compound undergoes an irreversible
one-electron electrochemical oxidation. Calculations have predicted this to be a radical cation of C2 geometry
with 60° end-to-end twist.
Molecules that present functional groups in an atypical
fashion, for example distorted from their preferred geometry,
are intriguing targets for synthesis. Examples of systems
previously examined include twisted amides1 and nitro substit-
uents2 and the highly distorted aromatic rings found in several
triphenylenes, naphthalenes, and pentacene cores.3-7 These
systems are inherently higher in energy than their non-strained
cousins and can be far more reactivesthe reactivity being driven
by release of steric strain.1,4 Overcrowded “D3h”-symmetric8
polycyclic aromatic hydrocarbons (PAHs) can adopt either a
C2 or D3 (propeller-like) conformation.9 The C2 versus D3
question is part of a wider issue regarding the driving forces
underlying the conformational preference, i.e., sterics versus
electronics, and considerable effort has been devoted toward
resolving this point. Triphenylenes are a family that fit under
the “D3h” umbrella. Currently, only the moderately twisted
peraryloxytriphenylene 1b10 and perfluorotriphenylene (2)3,6 and
the highly twisted perchlorotriphenylene (3)4,5b have been
studied crystallographically, while aryloxy compounds 1a,c and
perbicyclo[2.2.2]octenetriphenylene (4)11 have been synthesized
but not characterized by X-ray.12 However, Pascal et al. have
suggested that halogen substituents may themselves have
significant electronic effects which affect the structural char-
acteristics.9 As a result, they concluded that “the presently
available experimental structures are perhaps not the very best
examples for an examination of the C2/D3 dichotomy”. They
suggested and modeled four “ideal” candidates,13,14 one of which
was 1,4,5,8,9,12-hexamethyltriphenylene (5). They predicted
that 5 would adopt a C2 conformation, as opposed to D3, with
the C2 system being computationally more stable by >7 kcal/
mol. Therefore, the experimental realization of 5 would
provide a model system for investigating many of the key issues,
including conformational interconversionssan almost uncharted
area of knowledge for these hindered triphenylenes. We report
here the synthesis and complete characterization of compound
5. This characterization includes X-ray crystallography, variable-
temperature NMR (VT-NMR), molecular modeling, fluores-
cence spectroscopy, and cyclic voltammetry (CV).
(1) Tani, K.; Stoltz, B. M. Nature 2006, 441, 731-734.
(2) Hiyama, Y.; Brown, T. L. J. Phys. Chem. 1981, 85, 1698-1700.
(3) Hursthouse, M. B.; Smith, V. B.; Massey, A. G. J. Fluor. Chem. 1977, 10,
145-156.
(4) Shibata, K.; Kulkarni, A. A.; Ho, D. M.; Pascal, R. A., Jr. J. Org. Chem.
1995, 60, 428-434.
(10) Frampton, C. S.; MacNicol, D. D.; Rowan, S. J. J. Mol. Struct. 1997, 405,
169-178.
(11) Nishinaga, T.; Inoue, R.; Matsuura, A.; Komatsu, K. Org. Lett. 2002, 4,
1435-1438.
(5) (a) Pascal, R. A., Jr. Chem. ReV. 2006, 106, 4809-4819. (b) Shibata, K.;
Kulkarni, A. A.; Ho, D. M.; Pascal, R. A., Jr. J. Am. Chem. Soc. 1994,
116, 5983-5984.
(6) Smith, V. B.; Massey, A. G. Tetrahedron 1969, 25, 5495-5501.
(7) (a) Lu, J.; Ho, D. M.; Vogelaar, N. J.; Kraml, C. M.; Pascal, R. A., Jr. J.
Am. Chem. Soc. 2004, 126, 11168-11169. (b) Niemz, A.; Rotello, V. M.
Chem. ReV. 1999, 32, 44-52. (c) Kroto, H. W. Recent Res. DeV. Appl.
Phys. 2002, 5, 409-436. (d) Watson, M. D.; Fechtenko¨tter, A.; Mu¨llen,
K. Chem. ReV. 2001, 101, 1267-1300. (e) Wu, Y-T.; Siegel, J. S. Chem.
ReV. 2006, 106, 4843-4867. (f) Tsefrikas, V. M.; Scott, L. T. Chem. ReV.
2006, 106, 4868-4884.
(12) (a) Factual end-to-end twists: 1b, 45.2°; 2, 40.2°; 3, 56.6°. Predicted
(computed) twists: 4, 61°; unsubstituted triphenylene, 2°. (b) A 1:1 co-
crystal of perfluoro 2 and triphenylene gives corresponding twists of 33°
and 16°. Weck, M.; Dunn, A. R.; Matsumoto, K.; Coates, G. W.;
Lobkovsky, E. B.; Grubbs, R. H. Angew. Chem., Int. Ed. 1999, 38, 2741-
2745.
(13) The experimental structures investigated were hexabenzotriphenylene,
decacyclene, a hexa-tert-butyldecacyclene, perfluorotriphenylene (1), and
perchlorotriphenylene (2). The three remaining theoretical structures
examined, to the best of our knowledge, have not been synthesized. These
are a hexamethyldecacyclene and two isomeric hexafurotriphenylenes.
(14) Pen˜a, D.; Pe´rez, D.; Guitia´n, E.; Castedo, L. Eur. J. Org. Chem. 2003, 7,
1238-1243.
(8) The D3h symmetry is related to how molecules are drawn on paper. Often,
these structures are drawn without “wedged bonds”, which is indicative of
“D3h” symmetrysthis can be misleading.
(9) Barnett, L.; Ho, D. M.; Baldridge, K. K.; Pascal, R. A., Jr. J. Am. Chem.
Soc. 1999, 121, 727-733.
9
10.1021/ja074120j CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 13193-13200
13193

A R T I C L E S
Wang et al.
Scheme 1. Synthesis of 1,4,5,8,9,12-Hexamethyltriphenylene (5)a
a Reagents and conditions: (i) NBS, 15 °C, 78%; (ii) HMDS, 80 °C;
98%; (iii) BuLi, -100 °C; (iv) Tf2O, -100 °C, 66% (two steps); (v) CsF,
Pd(dba)2, 65%.
one methyl singlet (2.45 ppm; 18H). Initially, this was sugges-
tive of D3 symmetry, which was possible given the observations
by Guitia´n et al. and Bennet et al. that the D3 hexabenzotriph-
enylene product could be formed as the kinetic product21 under
similar mild reaction conditions. However, given Pascal’s
theoretical prediction that the C2 conformer would be the
thermodynamically most stable product, we decided to further
characterize 5 using VT-NMR, X-ray crystallography, and
computational modeling.
1
The results of the H VT-NMR experiment are shown in
Figure 1. On decreasing the temperature from 25 to -80 °C,
the parent peak associated with the aromatic protons (7.18 ppm;
25 °C) resolved completely into two broad singlets (7.21 and
7.05 ppm; 4H and 2H, respectively; -80 °C), with partial
resolution of the methyl singlet (2.45 ppm) into two overlapping
broad peaks (2.41 and 2.35 ppm). From these data, a coalescence
temperature (Tc) of -55 °C (220 K) was determined, which
reveals rapid room-temperature conformational interconversions
between two C2 enantiomers rather than the presence of one
D3 conformer in solution. The activation energy (Ea) associated
with this C2-C2 racemization was calculated using
Ea ) ∆Gq ) RTc[23 + ln(Tc/∆ν)]
where Tc ) 220 K and ∆ν ) 57.6 Hz. This gave Ea ) 10.20
kcal/mol, which is a relatively low activation barrier, given the
expected magnitude of methyl steric interactions. The rate of
this interconversion, k, at Tc, was determined by using the
formula
To prepare 5, we exploited the fact that arynes undergo metal-
catalyzed cyclotrimerization reactions to afford polycyclic
aromatics (Scheme 1).15-19 Treatment of commercially available
2,5-dimethylphenol (6) with N-bromosuccinimide (NBS; 15 °C)
and hexamethyldisilazane (HMDS) afforded bromosilyl ether
7. Sequential reactions with butyl lithium (BuLi) at -100 °C
and triflic anhydride (-100 °C) provided the key o-trimethyl-
silylaryl triflate 8. Then cesium fluoride was added in the
presence of Pd catalyst to give 5 in 65% yield (33% overall).20
k ) π∆ν/x2
as 128 s-1 22
. From the Arrhenius equation, assuming invariance
of the frequency factor with temperature between Tc and
(20) 1H NMR (250 MHz, CDCl3): δ ) 7.18 (s, 6H), 2.45 (s, 18H). 13C NMR
(62.5 MHz, CDCl3): δ ) 133.5 (C, C-a), 131.4 (C, C-b), 129.3 (CH, C-c),
22.9 (CH3, C-d). MS: m/z ) 312 (M+, 100), 154 (M - C12H12, 73), 77
(M - C18H19, 51). UV/vis (CH2Cl2): λmax (ꢀmax) ) 235 (12 500), 281
(23 200) nm (L mol-1 cm-1). FT-IR (cm-1): ν ) 3044, 2986, 2949, 2866,
1456, 1378. Elemental analysis, calculated: C, 92.24; H, 7.74. Actual: C,
92.26; H, 7.67. Mp ) 160-162 °C.
Examination of the NMR Traits of 5
1
The H NMR spectrum of 5 at 25 °C (Figure 1) revealed
only two sharp peaks, one aromatic singlet (7.18 ppm; 6H) and
(15) Pen˜a, D.; Pe´rez, D.; Guitia´n, E.; Castedo, L. Org. Lett. 1999, 1, 1555-
1557.
(21) (a) Pen˜a, D.; Pe´rez, D.; Guitia´n, E.; Castedo, L. Org. Lett. 1999, 1, 1555-
1557. Pen˜a, D.; Cobas, A.; Pe´rez, D.; Guitia´n, E.; Castedo, L. Org. Lett.
2000, 2, 1629-1632. (b) Bennett, M. A.; Kopp, M. R.; Wenger, E.; Willis,
A. C. J. Organomet. Chem. 2003, 667, 8-15.
(16) Pen˜a, D.; Cobas, A.; Pe´rez, D.; Guitia´n, E. Synthesis 2002, 1454-1458.
(17) Pen˜a, D.; Escudero, S.; Guitia´n, D. P. E.; Castedo, L. Angew. Chem., Int.
Ed. 1998, 37, 2659-2661.
(18) Pen˜a, D.; Pe´rez, D.; Guitia´n, E.; Castedo, L. J. Am. Chem. Soc. 1999, 121,
5827-5828.
(22) The formula for determining rates is more commonly used for decoalesced
peaks of equal intensity. However, the same formula can be applied to
peaks of unsymmetrical intensities, as seen in our VT-NMR studies. Kost,
D.; Carlson, E. H.; Raban, M. Chem. Commun. 1971, 66, 656-657.
(19) Pen˜a, D.; Pe´rez, D.; Guitia´n, E. Angew. Chem., Int. Ed. 2006, 45, 3579-
3581.
9
13194 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Flipping Twist in 1,4,5,8,9,12-Hexamethyltriphenylene
A R T I C L E S
Figure 1. (a) 1H VT-NMR spectra of 5 in 3:1 CS2/CD2Cl2 (360 MHz). Temperatures ranged from 25 to -80 °C. The aromatic signal at 7.18 ppm (25 °C)
resolved into two new signals, 7.21 and 7.05 ppm (-80 °C), and was concomitant with partial decoalescence of the methyl signal (2.45 ppm) into two
1
overlying peaks, 2.41 and 2.35 ppm. H NMR spectra were calibrated using TMS (0 ppm). Signal at 5.33 ppm is CH2Cl2. Note the “simple” NMR of 5 at
25 °C. (b) Inset showing scale-up of the aromatic region.
25 °C, this gives an estimate of k at 25 °C of 5.7 × 104 s-1
.
3; as a result, there are many parallels between the two systems.
First, triphenylene 5 takes on a conformation with C2 symmetry
with very large deviation from planarity. This severe distortion
enables alleviation of the “C:C” non-bonded contacts between
adjacent methyl substituents. Even so, the average of nine non-
bonded contacts in the crystal structure was 2.976 Å, well within
the sum of the van der Waals radii of two carbons (3.4 Å),
which indicates significant steric interaction. The CAr-CH3
radial bonds of 5 are all similar in length (the six radial distances
were 1.506, 1.513, 1.510, 1.507, 1.501, and 1.525 Å), with mean
1.510 Å. Second, rings B (C’s 9-10-12-13-14-16) and C
(C’s 17-18-20-21-22-24) adopt boat conformations (Figure
2) with distortions approaching those of [8]paracyclophane.5b
The remaining naphthalene substructure, rings A and D (C’s
24-17-16-9-8-1-2-4-5-6), exhibits a 53° end-to-end
twist, with the central ring A (C’s 24-17-16-9-8-1) and
ring D (C’s 8-1-2-4-5-6) contributing 35° and 18°,
respectively, to the overall twist. This value is similar to that
for 3, in which a 56.6° twist is present.12 Third, strong bond
1
This high interconversion rate explains the simple H NMR
spectrum that was observed at 25 °C.
It is interesting that only two distinct peaks were observed
at low temperatures in the 1H VT-NMR for the aromatic protons
in Figure 1, rather than the three expected for this compound
with C2 symmetry. This is due to an accidental isochronystwo
different types of protons with the same chemical shift (at 7.21
ppm at -80 °C). However, no accidental isochrony was
observed in 13C VT-NMR investigations at this temperature (see
Supporting Information). In fact, the three aromatic peaks
(132.900, 130.569, 128.771 (C-H) ppm) and one methyl peak
(22.257 ppm) at 25 °C decoalesced into nine aromatic (133.997,
132.950, 132.731, 131.987 (C-H), 131.199, 130.143, 128.975,
128.836 (C-H), and 125.894 (C-H) ppm) and three methyl
peaks (21.626, 22.734, 23.174 ppm) at -80 °C, respectively.
This provides additional support for a C2-symmetric conforma-
tion in solution.
X-ray Crystallographic Studies of 5
Unlike many other PAHs, compound 5 was found to be
reasonably soluble in organic solvents. Colorless crystals suitable
for X-ray studies were obtained by slow recrystallization of a
saturated solution in ethanol. The X-ray structure of 5 is shown
in Figure 2 in both framework and corresponding space-filled
representations.23,24 The crystal structure revealed a conforma-
tion similar to that of the highly twisted perchloro compound
(23) (a) Altomare, A.; Cascarano, G.; Giacovazzo, G.; Guagliardi, A.; Burla,
M. C.; Polidori, G.; Camalli, M. J. Appl. Crystallogr. 1994, 27, 435. (b)
Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.; Watkin, D. J.
J. Appl. Crystallogr. 2003, 36, 1487. (c) Spek, A. J. PLATON; University
of Utrecht: Utrecht, The Netherlands, 2006.
(24) (a) The crystal is a pseudo-merohedral twin. Parsons, S. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 2003, D59, 1995-2003. (b) The crystal structure
contains three crystallographically independent molecules. The majority
of molecular crystal structures form with one molecule in the asymmetric
unit. Steiner, T. Acta Crystallogr. 2000, B56, 673-676.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13195

A R T I C L E S
Wang et al.
tions. The predicted circumference was 8.72 Å, which matches
well with the observed circumference of 8.71 Å and the resulting
molecular shape. Second, it was anticipated that strong bond
alternation in the central ring A would exist, typical of a C2
conformer, and that exo and endo bonds with average lengths
of 1.491 and 1.417 Å, respectively, would be seen. These values
are reasonably close to the 1.478 (exo) and 1.425 Å (endo) bonds
lengths observed for 5. The difference appears to be due to an
overestimation of the distortion by these HF/3-21G(*) computa-
tions. Non-bonded contact distances were unavailable and only
low-level semiempirical (MNDO, AM1, PM3) and ab initio
Hartree-Fock (STO-3G, 3-21G(*)) C2 and D3 ground-state
computations without temperature correction were conducted
by Pascal for hexamethyl 5.9 We therefore decided to use density
functional theory (DFT) computational methods with enhanced
basis sets, which were expected to be more accurate, to examine
closely the ground states of the C2-5 and D3-5 conformers and
the proposed transition states for C2-C2 racemization and C2-
D3 interconversion at 298.15 K.
Molecular Modeling of Compound 5
Ground-state C2 and D3 structures were calculated using the
Gaussian 03 program.25 These stationary structures were
confirmed as energy minima by means of vibrational analysis
and the presence of zero imaginary vibrational frequencies. The
geometries obtained were fully optimized both using a lower
level ab initio HF/6-31G(d,p) calculation and an extended DFT
B3LYP level using the basis sets 6-31G(d,p) and 6-311G-
(d,p).26-29 Concomitant with the ground-state studies, the
transition states were initially located using semiempirical AM1
calculations implemented in Gaussian 03. These geometries were
then fully optimized and confirmed as energy maxima using
HF/6-31G and HF/6-31G(d,p), and B3LYP/6-31G(d,p) and
B3LYP/6-311G(d,p), functionals and basis sets. The fully
optimized structures for the C2-5 and D3-5 conformers, and the
proposed transition states for C2-C2 racemization (TS-I) and
C2-D3 interconversion (TS-II), are represented in Figure 4,
together with the corresponding wedge structures, which
demonstrate both the geometry and the symmetry of these
systems. The relative disposition of the six methyl groups can
be clearly seen.
The two transition states presented in Figure 4, TS-I and TS-
II, are shown using the B3LYP/6-311G(d,p) method and basis
set; they correspond to the C2-C2 and C2-D3 interconversions,
respectively. Looking closely at these structures, it can be seen
that TS-I presents “saddle-shaped” Cs symmetry, which results
in a relatively low activation barrier, while TS-II is a more
distorted C2 system and is consistent with a considerably higher
barrier to interconversion. The higher free energies calculated
for the D3 conformer (Table 1) explain the formation of the C2
as the most stable thermodynamic product. To confirm that the
C2 conformer was thermodynamically most stable, we heated a
Figure 2. Molecular structure of 5 at 150 K (top) and the corresponding
space-filled form (bottom). Hydrogen atoms shown in black. Thermal
ellipsoids have been drawn at 50% probability.
Figure 3. Structures showing ring labels and assignment of bond types
for 5. The wedge drawings correspond to the conformation present in the
crystal structure in Figure 2. Mean bond lengths (Å): endo, 1.425; exo,
1.478; non-bonded contacts, 2.976; radial, 1.510.
alternation is seen in the central ring A (Figure 3). The endo
bonds have mean length of 1.425 Å, while the exo bonds
average 1.478 Å. The existence of bond alternation means that
5 can be regarded as three benzene rings (B, C, and D) linked
together by bonds of low order. Indeed, it can be argued that it
is the drive for “conservation of aromatization” in rings B, C,
and D that induces such close C:C contact between the
neighboring methyl groups. This seems to be a common trait
in triphenylene compounds, such as triphenylene itself, perary-
loxy 1b, perfluoro 2, and perchloro 3. These properties lend
significant weight to the argument that the geometries of these
systems are governed by electronics and not sterics. Further
evidence supporting this dictum is provided by the electrostatic
potential maps of the C2 and D3 conformers of 5 (see Supporting
Information).
At first glance, the experimental data seem to fit with Pascal’s
computational predictions9 well. In particular, there are two
aspects of the comparison that can be examined more closely.
First, the circumference, which is measured as “3 exo + 3 endo
bond lengths”, arrived at from the computations can provide a
good indicator of the actual conformation. It was stated that
circumferences >8.6 Å would be associated with C2 conforma-
(25) Frisch, M. J.; et al. Gaussian 03, Revision C.02; Gaussian, Inc.: Walling-
ford, CT, 2004.
(26) The structures of all molecules were drawn using the visualization freeware
ArgusLab. Thompson, M. A. ArgusLab 4.0.1; Planaria Software LLC:
Seattle, WA, 2004 (http://www.arguslab.com).
(27) All calculations were run on the EaStCHEM research computing facility’s
Edinburgh computer cluster (http://www.eastchem.ac.uk/rcf).
(28) Output was viewed using the Gabedit visualization software, freely available
from the Internet (http://gabedit.sourceforge.net/).
(29) Images (Figure 4) were rendered using the Persistence of Vision Raytracer
(POV-Ray) freeware (http://www.povray.org).
9
13196 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Flipping Twist in 1,4,5,8,9,12-Hexamethyltriphenylene
A R T I C L E S
Figure 4. Schematic drawings of the C2 and D3 conformations and the transition states of hexamethyltriphenylene 5 (top). Corresponding optimized molecular
geometries for the B3LYP/6-311G(d,p) functional and basis set (bottom).
Table 1. Temperature-Corrected Calculated Free Energies of
Ground States and Barriers to Conformational Interconversions
6-311G(d,p), consistent with the fact that DFT takes into account
(Activation Energies) in kcal/mol at 298.15 K Relative to the C2
the B3LYP functional with two basis sets, 6-31G(d,p) and
electron correlation effects. Such differences are small and may
be due to the effects of solvation (not considered in these
calculations). All calculated C2-D3 energy barriers were similar
and significantly higher than the C2-C2 barriers.
Ground State
D3
C2
−
C2
C2−D3
method
conformer
TS-Ia
TS-II
AM1
HF/6-31G
HF/6-31G(d,p)
B3LYP/6-31G(d,p)
B3LYP/6-311G(d,p)
+1.41
+6.44
+5.91
+4.97
+4.97
+7.84
+11.11
+10.93
+9.80
+23.98
+30.40
+29.42
+25.42
+25.47
To ascertain the degree of fit between the computed C2
ground-state structures and the X-ray structure, root-mean-square
deviations (rmsd’s)30 were calculated using the Visual Molecular
Dynamics (VMD 1.8.5) program.31 These analyses revealed that
all the carbon skeletons of the modeled geometries matched
the corresponding backbone of the X-ray structures very well
(rmsd range 0.028-0.056), although some discrepancies were
observed. As expected, the B3LYP/6-31G(d,p) and B3LYP/6-
311G(d,p) produced C2 conformers with the best fit (mean 0.040
and 0.042, respectively). Furthermore, the non-bonded contacts
were best handled by these DFT methods; the average distance
for nine contacts in the crystal32 was 2.976 Å, while averages
of 2.996 (6-31G(d,p)) and 3.003 Å (6-311G(d,p)) were observed
for the DFT structures. The HF/6-31G and HF/6-31G(d,p)
geometries (mean 0.045 and 0.047, respectively) showed greater
deviation when compared to the DFT methods, which may be
due to the lack of electron correlation effects in the calculations.21a
However, the non-bonded contactss3.000 (6-31G) and 3.005
Å (6-31G(d,p))swere reasonably close to the experimental
value. The AM1 conformer (mean 0.045) seemed to provide a
match to the crystal data similar, overall, to HF; however, the
non-bonded contacts (2.908 Å) were grossly underestimateds
a problem that has been observed prior to this study.9 The exo
(1.478 Å) and endo (1.425 Å) bonds lengths, seen in the crystal
structure, were matched best utilizing the B3LYP/6-31G(d,p)
and B3LYP/6-311G(d,p) functionals and basis sets. These gave
corresponding mean lengths of 1.481 (exo) and 1.433 Å (endo)
and 1.479 (exo) and 1.431 Å (endo), respectively. Both of these
+9.64
a Exptl: 10.20 kcal/mol.
small quantity of hexamethyltriphenylene 5 in DMSO-d6 at
170 °C for 6 h. A carefully calibrated H NMR verified no
1
change in the spectrum. It is interesting that this synthetic
method has always previously given the C2 conformation as
the initially formed product, even when this is not the most
thermodynamically stablesas seen for the formation of the
kinetic C2 hexabenzotriphenylene product.21 The formation of
the thermodynamic product suggests that these synthesis condi-
tions favor the formation of a molecule with C2 symmetry,
perhaps because the reaction components come together in such
a way that the transition state closely resembles the C2 rather
than D3 structure.
Some deviation in the quantitative data in Table 1 is clearly
seen, as a result of the differing levels of theory employed for
the calculations. Semiempirical AM1 has been shown to be
fairly effective at modeling similar barriers observed in
hexabenzotriphenylene21a but underestimated the C2-C2 barrier
of 5, as is shown by this temperature-corrected data. Ab initio
Hartree-Fock calculations are known to overestimate energy
barriers to conformational interchange and, as expected, give
greater values for both C2-C2 and C2-D3 conformational
interchanges than the other methods. The variation of these
Hartree-Fock barriers could be the result of neglecting electron
correlation effects, which are anticipated to be significant for
the conformational interconversions that proceed via greatly
distorted transition states.21a The best agreement between the
calculated and experimental activation energies (within 0.4-
0.6 kcal/mol) was achieved by using DFT calculations using
(30) Due to the presence of poorly diffracting hydrogen atoms, only the carbon
backbones were compared.
(31) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graphics 1996, 14, 33-
38. The VMD 1.8.5 program is available from the Internet (http://
www.ks.uiuc.edu/Research/vmd/).
(32) All nine distances (Å): 3.007, 2.938, 2.967, 3.005, 2.985, 2.941, 2.955,
2.954, and 3.030.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13197

A R T I C L E S
Wang et al.
1
Figure 5. Wedge drawings of 5, illustrating the rapid conformational interconversions that provide averaged H NMR signals. Computations indicate that
the mechanism proceeds chiefly through successive low-barrier C2-C2 interchanges.
methods afforded extremely good fits and show the value of
employing DFT calculations for these systems. The HF and
AM1 methods gave poorer fits, both underestimating the degree
of distortion.
this. Perfluorotriphenylene (2), unlike 3, does have useful NMR
active nuclei for investigating the conformational dynamics in
solution. 19F NMR was performed only at 25 and -40 °C, with
the conclusion that “a propeller-like distortion (D3) must exist
in solution”,3,6 even though a C2 structure was observed in the
crystal (the thermodynamic product), and there was no sugges-
tion of conformational interchange. Since we believe an
incomplete picture for 2 persists, we performed AM1 and
B3LYP/6-311G(d,p) computational studies on this molecule.
Similar to our system, perfluorotriphenylene 2 can undergo
dynamic conformational interchange, although with a slightly
different mechanism. The 6-311G(d,p) C2-C2 and C2-D3
barrierss1.96 and 6.57 kcal/mol, respectivelysare both low.33
As a consequence, both the C2-C2 racemization and C2-D3
interconversions can occur very rapidly and easily at room
temperature, which gives rise to the appearance of a sharp
averaged “D3” NMR signal in solution, even at -40 °C. Neither
the peraryloxytriphenylenes 1a-c nor perbicyclo[2.2.2]-
octenetriphenylene (4) was investigated for any type of con-
formational interchange, as no low-temperature NMR studies
were reported. However, we anticipate that dynamic confor-
mational interconversions do occur for all these systems,
although to varying degrees. For aryloxy compound 1b, the
C-O inner bond length is 1.377 Å and the van der Waals radius
for oxygen is 1.4 Å.10 Given that these are slightly longer and
larger than the corresponding values for the C-F bonds of 2
(1.330 Å; vdW radius F ) 1.35 Å), it then seems reasonable to
imagine that dynamic conformational interchange is present and
occurs rapidly at room temperature, at a rate somewhere between
those observed for 2 and 5. Supporting evidence is provided
by the low-level AM1 calculation on a mimic of 4, which
suggests that the C2-C2 barrier is approximately 6.08 kcal/
mol.34,35 For peroctene 4, the steric demands are expected to
be greater than those for perchloro 3 and hexamethyl 5, since
molecular modeling studies predicted a 61° twist, although this
All calculations in Table 1 demonstrated that the C2-D3
energy barrier is substantially higher than the one for the C2-
C2 interconversion. An interesting upshot of this high barrier is
the implication that methyl flipping, between neighboring methyl
substituents, mostly occurs when the resulting conformer is C2,
not D3. By analyzing the conformational dynamics, a complete
picture for 5 is provided (Figure 5). Conformer A has three
possible sites for flipping: the neighboring methyl substituents
at the 1,12, the 4,5, and the 8,9 positions. First, flipping of the
8,9-methyl groups affords B, which is a D3 system, so it is
unfavorable. Second, the 4,5-methyl interchange is favorable,
as it provides C, which presents C2 symmetry. Third, the 1,-
12-exchange is also favorable, providing the C2 conformer D.
If this exchange is followed by an 8,9-flip, conformer E with
C2 symmetry results, meaning that the 8,9-flip is now favorable
(cf. A to B). In a short time, all the methyl groups are rapidly
scrambled (k ) 5.7 × 104 s-1), with the flipping mechanism
proceeding mainly through low-energy Cs transition states; rapid
flipping manifests itself as single sharp peaks in the NMR
spectrum (Figure 1a).
Conformational Analyses of Hindered Triphenylenes
1-4
This is the first time that solid evidence has been provided
for the dynamic conformational interconversion mechanism for
this class of hindered triphenylenes. Pascal’s group was the first
research group to propose possible dynamic exchange between
two enantiomers after synthesizing perchloro 3.4 Low-level AM1
calculations of this system were conducted after failing to
resolve the enantiomers by conventional means, and they hinted
at the possible existence of racemization that progressed via a
low-energy barrierswhere the calculated C2-C2 and C2-D3
barriers were 7.0 and 26.1 kcal/mol, respectively. However, due
to the physicochemical makeup of 3, it is not possible to validate
(33) The AM1 barriers were 0.82 and 3.45 kcal/mol for the C2-C2 and C2-D3
barriers, respectively.
9
13198 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Flipping Twist in 1,4,5,8,9,12-Hexamethyltriphenylene
A R T I C L E S
Figure 7. Cyclic voltammogram of 5. [5] ) 1 mM in background
electrolyte solution of 0.1 M anhydrous LiClO4 in acetonitrile (dried;
distilled). The reference electrode was made in-house and consists of a Ag
wire dipped into a solution of AgClO4 (0.01 M) in background electrolyte
solution, with a measured potential of -0.074 V vs the Fc/Fc+ couple. The
counter electrode was a 2 cm2 Pt gauze, and the working electrode was a
0.387 cm2 Pt rotating disc electrode. These were controlled by an
AUTOLAB PGSTAT30 potentiostat (Eco Chemie BV) equipped with GPES
4.9 software.
Figure 6. Excitation (at the peak emission wavelength, λem ) 430 nm,
left) and emission (at the peak excitation wavelength λex ) 280 nm, right)
spectra for 5. [5] ) 1 × 10-6 M in ethanol. Quantum yield (æF) was 7%
when measured with comparison to indole (æF ) 0.4). The emission peak
seen near 310 nm and excitation peak near 380 nm are both solvent Raman
bands. The quantum yield was estimated by making a comparative
measurement using indole as standard, which has a known quantum yield
of 40% in ethanol solution.36 Low (∼µM) comparable concentrations of
sample and standard were used to ensure that inner filter effects were
negligible. For both sample and standard, the quantum efficiency is
proportional to the ratio of the integrated emission intensity (the area under
the measured emission spectrum) to the integrated absorption intensity (the
area under the measured absorption spectrum).
Electrochemical Investigation of 5
Cyclic voltammetric (CV) analysis of hexamethyltriphenylene
(5; Figure 7) showed an irreversible one-electron oxidation wave
(Epa ) 1.10 V vs Fc/Fc+) at all sweep rates studied between 20
and 500 mV/s. This behavior is analogous to the previously
observed irreversible one-electron oxidation of the unsubstituted
triphenylene (Epa ) 1.50 V in benzonitrile vs Fc/Fc+).11 As
might be expected, consistent with the electron-donating nature
of the six methyl substituents, the oxidation peak of 5 was
intermediate between those of unsubstituted triphenylene and
peroctenetriphenylene (4), where 4 exhibited a reversible one-
electron oxidation wave (E1/2 ) 0.44 V vs Fc/Fc+). Komatsu
et al.11 stated that the dramatically decreased oxidation potential
of 4 was due to the intrinsic electronic effects of annelation
with the bicyclo units, which raise the HOMO level of the
π-system through the inductive electron donation and σ-π
interactions, coupled with the additional elevation of the HOMO
resulting from the large deviation from planarity. It is possible
that the electrochemical reversibility of 4 may be due to the
very bulky nature of the bicyclo units, which provide a
“protective shell” (akin to site isolation) surrounding the
oxidized species, preventing it from showing the enhanced
reactivity usually seen in these strained systems by undergoing
further chemical reaction to form redox-inactive species. This
would not be the case for 5 and triphenylene, for which
rearrangement and/or intramolecular coupling reactions are
likely. For this reason, it would be intriguing to synthesize and
electrochemically characterize a permethylated triphenylene,
which would be completely peripherally protected against
intramolecular coupling through complete functionalization with
12 methyl substituents.
not been verified by X-ray characterization. If correct, this
implies that that any conformational interconversions are likely
to proceed in a manner mechanistically similar to those for 3
and 5, although with higher activation barriers. A C2-C2
transition state was located at the AM1 level and produced a
barrier of 7.33 kcal/mol.35 Surprisingly, this was very close to
the barrier provided for 3 and 5, indicating rapid interconversion
at room temperature.
Fluorescence Spectroscopy of 5
Steady-state fluorescence spectroscopy showed that hexam-
ethyltriphenylene (5) is fluorescent, emitting light in the blue
region of the spectrum (Figure 6). Currently, the study of this
class of compounds is highly desirable due to their potential
applications in optoelectronics, such as in organic light-emitting
diodes (OLED).37 A large Stokes shift (∆λ ) 110 nm) between
the first absorption maximum (λmax ) 320 nm) and the emission
maximum (λmax ) 430 nm) was observed. This is consistent
with a relatively large change in molecular configuration upon
excitation. This is expected, as excitation should involve a π-π*
electronic transition, resulting in a decrease in π-bonding and
an increase in the importance of steric repulsion. The quantum
yield for fluorescence (æF) of 5 was measured as 7% at room
temperature. These spectral characteristics are comparable to
those reported for a dibenzotriphenylene derivative previously
studied, which also gives blue fluorescence and a similar
quantum yield.38
Calculations of the temperature-corrected free energies of 5
and the one-electron oxidation product 5+•, each with C2
symmetry, along with those for indole (In) and the one-electron
oxidation product of indole (In+•), were carried out using
B3LYP/6-311G(d,p) at 298.15 K in acetonitrile, using the
polarizable continuum model.39 This resulted in a calculated
oxidation potential of 5 of +1.11 V vs Ag/Ag+ (0.01 M),
(34) Due to the size and complexity of peraryloxytriphenylene (1b), a mimic
was employed in the modeling calculations. Six phenyl groups were
replaced with methyls at the 2, 3, 6, 7, 10, and 11 positions. However,
only the 1, 4, 5, 8, 9, and 12 positions are involved in the distortions and
were left unchanged. A complete structure of the mimic can be seen in the
Supporting Information.
(35) We assume that the underestimation that occurs for the AM1 barriers for
hexamethyl 5 also occurs for 1b and 4.
(36) Kirby, E. P.; Steiner, R. F. J. Phys. Chem. 1970, 74, 4480.
(37) Xu, Q.; Duong, H. M.; Wudl, F.; Yang, Y. Appl. Phys. Lett. 2004, 85,
3357-3359.
(38) Morrison, D. J.; Trefz, T. K.; Piers, W. E.; McDonald, R.; Parvez, M. J.
Org. Chem. 2005, 70, 5309-5312.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13199

A R T I C L E S
Wang et al.
Figure 8. Electrostatic potential maps for the optimized geometries of 5 and 5+•. Red is electron-rich, blue is electron-poor. As well as the overall decrease
in electron density upon oxidation (an overall shift in color from red to blue), the potential map for 5+• also reveals the decrease in π-bonding and delocalization
after oxidationsa blue electron-poor region is clearly seen where ring D is joined to ring A.
compared with the experimentally observed value of +1.14 V.
Given that equivalent calculations showed D3-5+• to be of
significantly higher energy than C2-5+•, this good agreement
between experimental and calculated oxidation potentials sup-
ports the formation of the more thermodynamically stable C2-
5+• as the oxidation product.
studies have revealed a highly distorted C2 structure with a 53°
end-to-end twist. We have shown through experiment and
computation that 5 displays dynamic conformational interchange
in solution, and we have provided computational evidence that
this phenomenon is active in all hindered triphenylene systems.
We have also shown it to be fluorescent in the blue and
electroactive, as well as possessing good solvent solubility and
electroactivity. Fluorescence, electrochemical measurements, and
calculations have shown that both the excited-state and oxidized
forms of 5 show significant differences in geometry, consistent
with the decrease in π-bonding brought about by excitation and
by oxidation, which increases the end-to-end twist to 60°.
The calculated structure of C2-5+• (Figure 8) shows an
increased twist in the naphthalene substructure, with rings A
and D (C’s 24-17-16-9-8-1-2-4-5-6) exhibiting a 60°
end-to-end twist. This is consistent with the calculated increase
in the methyl separations (C’s 3-23 and C’s 7-11) from 2.99
to 3.02 Å and can be attributed to an increase in the importance
of steric repulsion due to the reduction in aromaticity brought
about by removal of a π-electron. This increased twist also
appears to affect electronic delocalization (Figure 8). In 5,
electron density can be seen to be distributed across the entire
aromatic system. However, on oxidation, a region of lower
electron density (darker blue) can be seen where ring D joins
ring A (Figure 2), along with a region of higher electron density
(green area) in rings B and C. Significant bond lengthening is
also found in four of the six C-C bonds in ring D (see
Supporting Information). This is consistent with the increased
twist in ring D decreasing π-bonding and electronic delocal-
ization in and around this ring.
Acknowledgment. We thank the Royal Society for a Uni-
versity Research Fellowship (T.H.G.) and the University of
Edinburgh for Ph.D. studentships (A.D.S. and J.B.H.). This work
has made use of the resources provided by the EaSTCHEM
Research Computing Facility (http://www.eastchem.ac.uk/rcf).
This facility is partially supported by the eDIKT initiative (http://
www.edikt.org). We are also grateful to David A. Leigh,
Andrew Turner, Herbert Fruchtl, Ian Sadler, Anita Jones, and
John Miller for helpful discussions.
Supporting Information Available: Full descriptions for the
1
synthesis of 5, including H and 13C NMR, MS, UV, and FT-
In summary, the model compound 1,4,5,8,9,12-hexamethyl-
triphenylene (5) has been prepared for the first time. X-ray
IR spectra; X-ray characterization material, including data
collection and crystal parameters, atomic coordinates, anisotropic
displacement coefficients, bond lengths and bond angles, and
least-squares planes; a second cyclic voltammogram, and an
image of 5 fluorescing; electrostatic potential maps of C2 and
D3 conformers; rmsd table and calculated bond lengths for and
differences between 5 and 5+•; “absolute energies” and opti-
mized geometries (as Cartesian coordinates) for all calculated
structures; and complete ref 25. This material is available free
of charge via the Internet at http://pubs.acs.org.
(39) It has previously been shown that accurate calculation of absolute values
of experimental indole oxidation potentials is possible to within a few tens
of millivolts, making this suitable as a reference redox reaction. Calculation
of the free energy of the reaction 5+• + In f 5 + In+• gives the oxidation
potential of 5 relative to In. This can then be converted to a calculated
oxidation potential on any reference scale by using the experimentally
measured value of the indole oxidation potential. Experimental oxidation
potentials are obtained from the measured peak potentials, Ep, as Ep - E1/2
) 28.5 mV for a reversible one-electron redox reaction at 298 K, and the
half-wave potential E1/2 can be considered equal to the standard redox
potential (the oxidation potential) when reactants and products have similar
diffusion coefficients in solution. Kettle, L. J.; Bates, S. P.; Mount, A. R.
Phys. Chem. Chem. Phys. 2000, 2, 195-201.
JA074120J
9
13200 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007